INDEX

Explanations

information and knowledge-related terms, such as 'information', 'knowledge', 'facts', and 'advice'

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

alone

-0.71

 Parables

-0.66

igham

-0.65

awar

-0.64

rine

-0.64

gg

-0.62

kus

-0.62

mini

-0.61

neys

-0.61

atos

-0.60

POSITIVE LOGITS

 glean

1.08

arial

0.96

 retrieval

0.95

 overload

0.94

 gathered

0.94

 regarding

0.92

 pertaining

0.89

 about

0.87

 theoret

0.86

 gathering

0.84

Activations Density 0.471%

information and knowledge-related terms, such as 'information', 'knowledge', 'facts', and 'advice'

No Comments

No Known Activations